The aims of the review were clearly stated for participants, treatment, study design and outcomes. The search covered a number of sources, but was limited to English language; it was possible that language bias or publication bias may have affected the review. The methods of data extraction were those aimed at reducing reviewer error or bias, but methods for study selection and quality assessment were unclear.
Quality assessment of included trials was limited. Given the lack of significant heterogeneity between trials, the methods of synthesis appeared appropriate. However, the authors gave little information about differences between trials, in particular any adjunctive drug therapies, or whether comparators in the individual trials were controls or placebo. Details of the participants were also limited; this could affect the generalisability of the results. There was no mention of the assessment of any possible adverse effects. There were discrepancies between some of the numbers in tables and text.
Most of the data came from three older trials (1975 and 1990). Drug regimes and lifestyles have changed over time, in particular with the use of statins, and results may not now be as relevant to current practice. The authors' conclusions should be treated with caution.